Abstract: Introduction
Laboratory studies of performance limitations under dual-task conditions typically present two tasks—Task1 (T1) and Task2 (T2) —in rapid succession. The degree of task overlap is varied by manipulating the interval between the onsets of the T1 stimulus and T2 stimulus, known as the stimulus onset asynchrony (SOA). The standard finding is that RT for T1 (RT1) is constant across SOA, but RT for T2 (RT2) increases dramatically as SOA decreases. This slowing of RT2 at short SOA is commonly known as the psychological refractory period (PRP) effect. The most widely accepted account of the PRP effect is the response-selection bottleneck (RSB) model, which assumes that central operations (such as response selection) are discrete and serial, central operations for T2 do not start until central operations for T1 have finished. According to this model, RT2 is delayed at short SOA because operations for T2 requiring the bottleneck mechanism have to wait until this mechanism is finished with T1. The longer SOA is, the less likely is processing T2 in the bottleneck to have to await the completion of processing T1 in it. Hence, this model makes rather simple predictions for RT1, because T1 gets access to bottleneck processing first, it has completed bottleneck processing by the time T2 requires it, or T2 has to wait for T1 until which has finished with the bottleneck. Regardless, T1 always gets access to the bottleneck as soon as it requires it, so RT1 should be the same at all SOA, also RT1 should not depend on any difficulties of T2. Thus, the present study aimed to test whether T1 response selection could be affected from T2 response selection information. If it is true, it would suggest that T2 response selection had begun before T1 response selection was completed—that is, response selection processes of both tasks were operating in parallel.
Method
Three reaction time experiments using a psychological refractory period paradigm examined whether the mental rotation process could occur in parallel with other cognitive operations. In each experiment, participants made speeded responses to both a tone (T1) and a different rotation letter or digit (T2), which presented with varying stimulus onset asynchronies. The first task was high-low tone discrimination, and the second task was a mirror-normal judgment. The tone stimulus was always presented first, and the rotation stimulus appeared after a variable SOA, which was 50ms or 500ms.
Results
The results revealed that: (1) T1 response selection heavily affected T2 response selection, the effect of PRP was significant on T2. The effect of mental rotation decreased substantially with SOA decreasing. (2) A significant effect of SOA was observed on T1. As SOA decreased, RT1 increased. T1 was significantly influenced by the difficulty of T2. (3) Response selection of T2 substantially affected the central bottleneck process of T1, such effect suggested that T2 response information was generated prior to the completion of T1 response selection, mental rotation process can occurs in parallel with other cognitive operations.
Conclusions
The present results suggested that T1 and T2 response selection processes may operate in parallel. Such parallel operation of response selection processes would be a violation of the discrete-stage processing assumption and could pose a substantial challenge to RSB theory

Key words: psychological refractory period, mental rotation, serial process, parallel process